One of the main difficulties today in ophthalmology is the lack of a good tomographic device which will give an instant picture of large parts of the retina and provide depth information at the same time. There are cameras which scan through the retina and there are cameras which take one or more images. Thus the choice of the ophthalmologist is between having good depth information in a few and sometimes unknown locations or a superficial picture of larger areas of the retina, but without depth information.
Such information is necessary in order to prevent some of the most blinding diseases, if only detected early enough. The prevalence of retinal diseases is significant among the older population, and early discovery of these diseases allows treating them in time, thus preventing full blindness of the patients. This early discovery is difficult because of a few reasons. Firstly, the optics of the eye hinder detection of details smaller than ten microns across, and usually even much coarser details are not seen. While adaptive optics, in the future, might partially solve this problem, it is still limited by technology and by availability.
Even if the optics problems are solved, one still has to measure the retina in depth, somewhat as in microscopy. This means getting three-dimensional pictures of the retina.
Prior art instruments which provide three-dimensional images of the retina are well-known by those skilled in the art. Current and commercially available devices are slit lamps, Optical Coherence Tomography (OCT) devices and confocal Scanning Laser Ophthalmoscopes (cSLO). They are all described in medical and optical literature.
Unfortunately different schemes to get these depth pictures require some kind of scanning, which leads to other problems. As the area and depth are divided into more resolution elements, it becomes nearly impossible to accumulate them all in a single picture. Simple slit lamp scanning, OCT, and cSLO (confocal microscopy) go sequentially from one point in the retina to the next, which must take time. Unfortunately, the patient's eye is a live object and it moves during scanning.
In the slit lamp, for example, a sheet of light illuminates a curve across the retina, and a camera is used for looking from the side, measuring the shape and intensity of the imaged light. The process is sequentially repeated by illuminating one slice after another of the retina. The major drawback is the fact that the eye frequently moves between these scans. This means that the separate scan results may no longer be ordered in space as they were intended, and special efforts must be taken to fit them into their original locations and to produce a composite full image.
In Optical Coherence Tomography (OCT), there is also a three dimensional scan, in two lateral directions and in depth. In newer versions some of these scans are replaced by direct imaging of the whole surface or by Fourier scanning. In OCT depth resolution is the best, but the scanning is slow. As a result, the emerging picture is not continuous. For cases where following the disease over time is essential, it is sometimes next to impossible to scan the same region again.
In confocal microscopy—i.e., using confocal Scanning Laser Ophthalmoscopes (cSLO)—a single beam is focused at a location in the retina, and the reflected image is scanned in depth through a pin-hole simultaneous with the beam, providing depth information at this point. The process is repeated at the next spot of light, and again it is hoped that the eye does not move between these scans. In some applications, multiple foci are provided simultaneously, and scattered light is imaged through a corresponding array of pinholes.
U.S. Pat. No. 4,883,061 to Zeimer, and U.S. Pat. No. 6,267,477 to Karpol and Zeimer disclose imaging scanning apparatuses and methods for retinal thickness non-invasive analysis. They include a light source, separate or common focusing optic and beam deflector for incident and reflected beams going to and returning from the retina of an eye, and an imaging device. The apparatus further includes separate optical paths for imaging the fundus and iris of the eye. An eye model is obtained by spatially integrating images of the retina, the fundus and the iris.
Specifically, Karpol and Zeimer in U.S. Pat. No. 6,267,477 teach using a single scanning laser beam for measuring retinal thickness while “ . . . changing the beam profile so that the subsequent profile of the beam at the retina is slit shaped and not dot shaped.”
To parallelize the scan, Basu and Moore, in Canadian Application CIPO 2284299, scan an array of lines or dots across the retina, or alternatively let the eye scan across such an array, and combine their results into a three-dimensional map of the retina. Similarly, Verdooner et al in U.S. Pat. No. 5,220,360, rotate a grid of parallel lines to fall on the retina at cross directions. The different line images are narrowed or skeletonized in software, as is known to those skilled in the art, to create a retinal topographic map, essentially assuming a single surface reflection.
Why is retinal slit scan preferred? Milbocker and Reznichenko, in Applied Optics, Vol. 30, p. 4148 (1991), prove mathematically and experimentally that having the beam arrive from one side of the pupil, and depart from the opposite side, has the best depth resolution. This resolution is more than double compared to methods where the beams arrive from the side and leave from the pupil center, or the other way round.
In the Journal of the Optical Society of America A, vol. 24, p. 1295 (2007), Wanek, Mori and Shahidi say about a system equivalent to that of Karpol and Zeimer: “We have developed an optical section retinal imaging technique with high spatial resolution and a depth resolution intermediate between that of SLO and OCT. Though our technique has lower depth resolution than OCT, it offers the advantages of higher image acquisition rate, better coverage for retinal thickness mapping, and flexibility in varying the incident laser wavelength to image fluorescence as well as reflectance.” Nevertheless, this is still a scanning technique.
The limitation of all scanning methods is that later data processing tries to remove the discontinuities in the scanned volumes. This is important also if one wishes to identify and track retinal positions for later follow-ups. All of these problems are much easier if the whole retina is measured in one take, without the need for repetitive scans and measurements. In this case, there is no ocular movement, no need to stitch the scans back together into the image, and no attendant distortion in the result.
In WIPO application WO2006/030413 to Iddan et al a large area of the retina is illuminated, being imaged through a Hartmann-Shack wave front sensor. This is performed without scanning. However, since the light source (namely the retina) in this prior art device is extended, the sensor fails to provide directional information about the returned beam.
The example of the retina is brought because of the many constraints encountered in both illumination and detection through the pupil, in the presence of aberrations which distort both incoming and outgoing beams. There are different, usually easier, constraints in other applications such as visualizing the depth information of nearly-transparent objects, such as plastics, gels, smoke or steam, biological in-vivo and in-vitro samples and more. If there is some light scattered within these objects, it can be used to trace their internal structure in three dimensions. In other cases, fluorescence is used, then again in others non-linear effects, one or more of these mechanisms serving to illuminate the light path when observed in directions other than the original one. Here, and hereinafter, these different mechanisms and processes are referred to in the general name of scattering, without losing generality.
In microscopy and other applications, one uses structured light: a scanned sheet of light, or a set of such sheets, which illuminate the sample. Light scattered from the sample along these sheets is then detected in a different aspect angle. Deviations and intensity variations in the scattered image of the sheet are translated into positional information and scatterer density. Stroboscopic methods allow separation of the sheets in time and thus in space. If the sheets are all illuminated at the same time, they must be separated such that the light returned from each of them does not become mixed-up with light scattered from others. This is the “venetian blind effect”, described by Ribak and Ragazzoni in the Proceedings of the European Southern Observatory, vol. 58, p. 281 (2001). The venetian blind effect imposes strict limitations on the mutual spacing of the sheets or on separate light beams, in the sense that the light scattered from separate sheets or foci does not fall on the same detector pixel.
In some of the prior art applications the time factor is important, such as in those involving smoke or steam turbulent motion, and scanning the structured light is too slow. In some other examples one wishes to avoid moving components and successive illumination of the light sheets or foci, and in others, speed of measurement is important. So concentrating on the more limiting example of retinal imaging does not limit the present invention only to this example; rather it serves as a descriptive means.
Thus there is a need for a system and a method to enable capturing a single-shot picture of a three-dimensional object, such as the retina of the eye or a non-ocular object, taken during a time when the object is immobile to provide depth information about the object.